AI User Research Analyst
An AI User Research Analyst specializes in studying human interactions with AI-powered products to generate actionable insights th…
Skill Guide
Usability Testing for AI Products is the systematic evaluation of an AI system's effectiveness, efficiency, and user satisfaction by observing real users performing tasks with it, specifically focusing on the unique challenges of AI-driven behavior, transparency, and user trust.
Scenario
You are tasked with testing a new AI tool that drafts email replies based on a few bullet points. Users are reporting that the tone is often wrong and they spend time rewriting.
Scenario
A healthcare chatbot asks symptom questions and suggests possible conditions. It sometimes provides an incorrect triage level (e.g., suggesting 'emergency' for a common cold). You need to assess if users can identify and recover from such errors.
Scenario
A streaming service's AI recommendation engine has high click-through rates but low long-term user satisfaction, suggesting filter bubbles. Leadership wants to improve the diversity and serendipity of recommendations without hurting engagement metrics.
Use UserTesting or Lookback for moderated remote sessions with screen/face recording. Maze is excellent for creating unmoderated, task-based tests with AI interaction flows. Hotjar provides heatmaps and session recordings to see how users actually interact with AI UI components.
The HAX Toolkit (Microsoft) provides design guidelines and test scripts for common AI patterns. The Transparency Checklist ensures you test for explainability. The Trust Calibration Framework helps measure if users appropriately rely on the AI based on its actual competence.
Adapt SUS questions to include AI trust (e.g., 'I felt confident using this AI'). UCE measures the gap between a user's trust in the AI and the AI's actual accuracy. Task-Specific Error Rate tracks failures unique to AI (e.g., 'prompt misinterpretation' vs. 'click error').
1 career found
Try a different search term.